DE102006026945A1 - Computer tomographic imaging method, involves irradiating object to be examined with polychromatic X ray and calculating location-dependent spectral absorption coefficient of object by reconstruction algorithm on basis of projection images - Google Patents

Abstract

The method involves irradiating an object to be examined (10) with polychromatic X ray (R) along a number of projection vectors (d). The projection images (P) of the X ray transmitted through the object are recorded. A location dependent spectral absorption coefficient (K(r,E)) of the object to be examined is calculated by a reconstruction algorithm on the basis of projection images. The transmitted X ray energy drained is detected. A location dependent spectral absorption coefficient is calculated, by separate reconstruction of quasi-monoenergetic groups of projection images (Pi). Independent claims are also included for the following: (1) a method for determining the location dependent concentration of a number of materials of the object to be examined (2) a computer tomograph for the execution of the imaging method.

Description

The The invention relates to a computed tomographic image acquisition method as well as a computer tomograph working according to this method, in particular for use in the field of medical technology. The invention further relates to a method for determining the location-dependent concentration a number of predetermined substances in a study object, the the result of the aforementioned image recording method as Input quantity.

In the conventional one Computed tomography are under irradiation of an examination subject with X-rays by means of an X-ray detector Projection images taken along different projection vectors. Each projection image gives an intensity distribution of the X-radiation again, along the associated Projection vector transmitted through the object to be examined and this is more or less mitigated. The while an examination of the examination object taken projection images are collectively referred to as sinogram.

By means of reconstruction (or back projection), the location-dependent (X-ray) attenuation coefficient of the examination subject within the area spanned by the projection vectors is calculated numerically from the sinogram. The spatial dependence of the attenuation coefficient in the examination object results in a contrast which visually reproduces internal structures of the examination object as a function of the space r (the underscore denotes the vectorial property of each spatial point). This image information is called a tomogram.

In the medical technology application is the object under investigation around the body a patient or part of it.

to Recording the projection images is usually polychromatic X-rays used, especially ordinary X-ray tubes of house emit polychromatically. As polychromatic X-ray becomes x-ray radiation denoting the different frequency components with the one by one more or less broad spectrum predetermined frequency contains. Instead of the x - ray frequency ν is in For the purposes of this application also to the quantum energy E of the X-ray radiation Referenced. These sizes are equivalent to each other and can by the known physical conversion rule E = h · ν into each other be converted. The letter h stands for the Plancksche Constant.

The commonly used X-ray detectors are insensitive to the different quantum energy of the X-radiation. The projection images thus only contain information about the total radiation incident on a particular projection vector at each pixel of the detector. Due to the measuring principle, however, the quantum energy of this radiation is averaged. During the subsequent reconstruction, a weakening coefficient μ ( r ) is derived from these projection images, which, although spatially resolved, allows an attenuation value to be assigned to a spatial point of the examination object, which is likewise integrated via the quantum energy of the x-ray radiation.

In fact, however, the X-ray attenuation of various body substances depends in different ways on the quantum energy. The energy-averaged attenuation coefficient μ ( r ) is thus a generic quantity that more or less falsifies the actual X-ray attenuation in a patient's body. The integrating property of conventional computed tomography leads in particular to dense absorbers, such as bone or metal, to artifacts (eg the so-called beam hardening), which impair the usability of the resulting tomogram.

The conventional Computed tomography also has only a limited contrast resolution that the Differentiation of various soft tissues (e.g., liver tissue, white and gray brain mass, etc.) sometimes makes it difficult or impossible.

to solution These problems are offered by CT scanners in which the tube voltage the X-ray source, and over here the spectral distribution of the emitted X-ray radiation adjustable is. With such devices become the projection images several times, each with different chromatic X-rays added. In an alternative embodiment of such a device two mutually crossed image systems each with an X-ray source and a detector used, wherein the X-ray sources of the two image systems with different tube voltage be operated and, accordingly, turn a different chromatic X-rays emit. However, even with these methods for each of the both X-ray spectral distributions the associated Sinogram energy-averaged recorded.

Before or in the course of the reconstruction, these sinograms are then linked by means of correction algorithms, in particular base material decomposition or Rho-Z projection, in such a way that the abovementioned artifacts are at least attenuated become. An example of such a correction algorithm is described in A. Macovski, et al., Energy Dependent Reconstruction in X-ray Computerized Tomography, Comp. Biol. Med. 6 (4): 325-336 (1976).

Alternatively, energy-selective detectors for computed tomography are currently being developed, which detect the transmitted X-radiation in each pixel selectively in accordance with a number of predetermined energy channels. A direct-converting energy-sensitive detector is for example from the DE 102 12 638 A1 known. On the other hand, energy-selective scintillator-based X-ray detectors are already being used in the framework of other imaging medical procedures, in particular PET or SPECT.

A method for the spatially resolved determination of the element concentration in examination objects is further from the DE 103 52 013 A1 known. Thereafter, to determine the respective concentration of N (N = 2, 3,...) Elements or combinations of elements, M (M = 3, 4,...) X-ray images each having a different spectral distribution are recorded under the same geometric conditions and for the respective spectral distribution to the location-dependent, energy-averaged attenuation coefficient μ _i ( r ) (i = 1,2, ..., M) to calculate. Taking into account the attenuation spectra of the sought-after elements or element combinations, their respective concentrations are then determined from these variables by solving a linear system of equations.

Of the Invention is based on the object, one before the above Background improved computer tomographic image acquisition method and a suitable computer tomograph for performing this Specify method. The invention is still the task based on the image recording method a simple and precise Method for determining the concentration of a number of predetermined Specify substances in a test object.

Regarding the Image recording method, the object is achieved by The features of claim 1. The invention is based on a computed tomography Method in which detects the transmitted X-ray energy resolved becomes. According to the invention resulting sinogram in groups of quasi-monoenergetic projection images structured, so almost exclusively on X-ray radiation of a common Quantum energy go back. As quasi-monoenergetic radiation with quantum energy is within an energy interval, within the sen the spectral attenuation coefficient of body substances approximately is constant. The width of such an energy interval is preferably between 0.01 keV and 80 keV and can also be used with the medium energy of the Energy channels vary to the generally non-linear energy dependency of the spectral scum coefficients to take into account.

The reconstruction is carried out separately for each of these groups of quasi-monoenergetic projection images. In this way, it is achieved that the reconstruction no longer computes the energy-averaged attenuation coefficient μ ( r ) but directly a location-dependent spectral attenuation coefficient κ ( r , E). The tomograms based on this spectral attenuation coefficient κ ( r , E) are inherently not subject to the above-described artifacts due to the omitted averaging process.

For the procedure can advantageously the common ones Reconstruction algorithms without significant adaptation effort be used. Especially as the occurrence of the averaging artifacts described above inherently avoided, unnecessary They themselves known correction algorithms. Especially with the use of this Procedure always also some deterioration of data quality, in particular the signal-to-noise ratio, is connected by the present method against the usual Procedures that rely on such correction algorithms had, with the same x-ray dose achieved an improvement in data quality. Alternatively to this Due to the improved recording technology a consistent data quality can already at a lower x-ray dose be achieved.

In order to increase the contrast of the tomogram, the spectral attenuation coefficient is again integrated via the energy according to its calculation. The result of this integration provides the same size, which is also determined in the course of conventional computed tomography, namely the location-dependent energy-averaged attenuation coefficient μ ( r ). However, since the integration step according to the invention takes place only after the reconstruction step, this size is obtained without the conventionally present artifacts or the quality loss caused by conventional correction methods.

In order in particular to increase the contrast caused by different types of soft tissue, in particular a weighted integration with the negative cube of the energy, ie with E ^-3 , of the spectral attenuation coefficient κ ( r , E) has proven to be advantageous.

With respect to the method for the location-dependent determination of substance concentrations in the examination subject, the object according to the invention solved by the features of claim 4. Thereafter, the orthogonal spectral attenuation coefficient κ ( r , E) is used as input. The concentration is determined by adapting to this input variable by means of a regression algorithm a model function containing as fixed parameters the attenuation spectra λ _i (E) of the given substances and as free parameters the location-dependent concentrations of these substances.

The use of the spectral attenuation coefficient κ ( r , E) as the input quantity of the concentration determination makes it possible to considerably simplify the mathematical method for concentration determination, since this variable is free from the complex interaction of the energy averaging with the downstream reconstruction, which is the cause of the described artifacts. Rather, the fact that the spectral attenuation coefficient κ ( r , E) provides information about the energy distribution of the X-ray attenuation for each space vector r enables a direct comparison of this quantity with the characteristic attenuation spectra of these substances, which are known or at least empirically easily determinable as material constants ,

The application of a regression method (also referred to as "fitting") thereby represents a particularly easy-to-handle, and at the same time precise, possibility for carrying out this comparison. A fixed parameter of the model function used in the regression is a parameter which is given in FIG the adjustment is kept constant. A free parameter is accordingly a parameter which is varied during the adaptation. The regression algorithm used is, in particular, the Levenberg-Marquardt algorithm known per se. In a particularly simple and numerically stable realization of the method, the model fiction is formed in particular as a linear combination of the attenuation spectra λ _i (E) of the substances to be determined.

In a preferred embodiment of the method, the substances to be determined comprise at least one element occurring in the body, in particular hydrogen, oxygen and / or calcium. The model function accordingly contains the substance-specific (X-ray) attenuation spectrum λ _i (E) of this element as a fixed parameter.

Alternatively or additionally, the substances to be determined comprise at least one homogeneous body substance, in particular water, fat, soft tissue or bone, wherein the model function contains the particular empirically determined attenuation spectrum λ _i (E) of this body substance.

With the spectral attenuation coefficient κ ( r , E), the concentration determination method uses the output of the above-described image pickup method and is closely related thereto. Preferably, both methods are also used in combination with each other. Nevertheless, the concentration determination method can also be used independently of the image recording method, in particular to further process otherwise generated and already existing or generically generated data sets of the spectral attenuation coefficient κ ( r , E).

With regard to the computer tomograph, the above object is achieved according to the invention by the features of claim 9. The computer tomograph then comprises an (X-ray) detector, which is designed for energy-resolved detection of the transmitted X-radiation and a reconstruction unit, which is designed to quasi by separate processing monoenergetic groups of projection images to calculate a location-dependent spectral attenuation coefficient κ ( r , E).

In a first variant of the invention, the detector is designed as a direct converter with energy-sensitive quantum counting electronics. With regard to the concrete realization of such a detector is in particular on the DE 102 12 638 A1 Referenced.

In According to a second variant of the invention, the detector comprises at least a scintillator layer with a downstream photosensor and a Quantenzählelektronik on. The detector is in particular with a sandwich arrangement provided by scintillator layers. Such detectors are in particular already used in PET or SPECT.

Preferably, the computed tomography device is simultaneously designed to carry out the concentration determination method described above. For this purpose, it comprises a concentration determination module in which the regression algorithm described above is implemented, and to which the spectral attenuation coefficient κ ( r , E) determined by the reconstruction unit is supplied as an input variable.

following is an embodiment of Invention explained in more detail with reference to a drawing. In it shows the only one Figure in a schematic block diagram of an inventively designed CT scanners.

The figure shows in a schematically simplistic simplification a medical computer tomograph graph 1 , The computer tomograph 1 includes an image system 2 with an X-ray tube as X-ray source 3 and an (X-ray) detector 4 , The computer tomograph 1 also includes a computer system 5 and peripherals 6 for data entry, output and archiving.

The X-ray source 3 and the detector 4 are in juxtaposition to one another around a common isocentric axis 7 rotatably mounted. For common storage of the X-ray source 3 and the detector 4 In particular, a gantry (not explicitly shown) is provided in a conventional computer tomography or a C-arm.

The computer system 5 includes a reconstruction unit 8th , in particular in the form of a software module in which a numerical reconstruction algorithm is implemented. The computer system 5 further comprises a concentration determination module 9 ,

The peripherals 6 For data input and output, conventional means such as screen, keyboard, pointing device (ie, mouse, trackball, etc.) and printer. The peripherals 6 further comprise an interface to a communications network for communicating with external devices and users. For archiving the examination data, a data memory and / or a network access to such are provided.

In the operation of the computer tomograph 1 becomes the body of a patient as an object of examination 10 approximately centered with respect to the isocentric axis 7 between the X-ray source 3 and the detector 4 positioned. The examination object 10 becomes successive rotation of the image system 2 around the isocentric axis 7 from different projection direction, and thus along different projection vectors d , irradiated with polychromatic X-ray radiation. For each projection vector d (the underscore stands in turn as a mathematical symbol for the vectorial character of the quantity d ) is determined by means of the detector 4 a projection image P by the examination object 10 transmitted X-rays R and to the reconstruction unit 8th transmitted. Based on the projection images P calculates the reconstruction unit 8th subsequently (in a manner described in more detail below) the location-dependent spectral attenuation coefficient κ ( r , E) of the examination subject 10 for each space vector r within an area spanned by the projection vectors d and as a function of the quantum energy E of the x-ray radiation R. The location dependence of the attenuation coefficient κ ( r , E) in the examination subject 10 gives a contrast, the internal structures of the examination subject 10 as a function of the space vector r . The attenuation coefficient κ ( r , E) can be given for a given quantum energy E as a tomogram in two- or three-dimensional spatial representation via the peripheral devices 6 are displayed.

To improve the contrast, the spectral attenuation coefficient κ ( r , E _i ) can alternatively be integrated after the reconstruction weighted by the quantum energy E, the resulting size μ ( r ) = ∫w (E) · κ ( r , E) dE, instead of the spectral attenuation coefficient κ ( r , E _i ) is displayed as a tomogram. The quantity μ ( r ) is referred to below as the energy-averaged attenuation coefficient. In particular, the weighting factor w (E) becomes w (E) = E -3 used.

The detector 4 includes along a the X-ray source 3 facing detector surface 11 a one- or two-dimensional array of detector elements 12 , Each detector element 12 corresponds to a pixel of the detector 4 generated projection images P. The position of each detector element 12 within the detector surface 11 and correspondingly the position of the associated pixel within the projection image P, are described by a coordinate vector x . The detector 4 is after the in DE 102 12 638 A1 built principle described. Alternatively, a scintillator-based sandwich detector such as commonly used in PET or SPECT is used.

Each detector element 12 is adapted to selectively register each incident X-ray quantum for the corresponding quantum energy E in one of N energy channels E _i = {E ₁ , E ₂ ,..., E _N }, so that in each energy channel E _i exclusively those of the detector element 12 are counted incident X-ray quanta whose quantum energy E in an interval [E _i - Δ / 2; E _i + Δ / 2]. The letter N stands for the total number of energy channels E _i . The letter Δ stands for the energy difference between two adjacent energy channels E _i and thus also for the energy interval covered by each energy channel E _i . The number N of the energy channels E _i is so high and the energy difference Δ chosen so low that the counted in each energy channel E _i X-ray quanta in the sense of the above definition can be regarded as quasi-monoenergetic. The detector 4 comprises in a preferred embodiment per detector element 12 N = 5 energy channels with an energy difference or an energy interval of Δ = 22 keV each.

The detector 4 Thus, for each projection vector d as the projection image P, the number of registered X-ray quanta as a function of the coordinate vector x and of the energy channel E _{i is given} to the reconstruction unit 8th out: P = P i ( d . x , E i ) with i = 1, 2, ..., N.

In the reconstruction unit 8th the projection images P are divided into groups, each of which contains the projection images P _i ( d , x , E _i ) associated with a common energy channel E _i . For these quasi-monoenergetic groups of projection images P _i ( d , x , E _i ), the reconstruction unit leads 8th the reconstruction in each case separately, wherein preferably one of the conventional computer tomography conventional reconstruction algorithms is used.

Due to the use of quasi-mononergetic projection images P _i ( d , x , E _i ) as the input, the reconstruction unit supplies 8th for each energy channel E _i the location-dependent, spectral attenuation coefficient κ ( r , E _i ) as a result.

Additionally or alternatively to the display of the tomogram on the peripherals 6 is determined by the concentration determination module 9 the spatially resolved concentration c _j ( r ) (with j = 1, 2, ..., M) of a number of M substances, eg the elements hydrogen, oxygen and calcium, in the object under investigation 10 certainly.

For this purpose which, as described above, certain spectral attenuation coefficient κ (r, E _i) used as input variable for a regression method, wherein for each spatial vector r a model function F (r, E _i) κ of spectral attenuation coefficients (r, E _i) is adjusted. The model function has the form F ( r , E i ) = Σ j c * j ( r ) · Λ j (e i ) + R ( r , E i ) and contains in linear combination the known attenuation spectra λ _j (E) of the substances sought, in particular therefore the elements hydrogen, oxygen and calcium for the values of the energy channels E _i . In the course of the regression, the attenuation spectra λ _j (E _i ) are treated as fixed, ie unchangeable, parameters. The linear coefficients c * _j ( r ) are varied as free parameters. The function R ( r , E _i ) is a residual term which, after completing the fit, ie for F ( r , E _i ) = κ ( r , E _i ), is a measure of the quality of the fit result and, at best, only Contains noise.

After successful adaptation give the linear coefficient c * _j (r) to a good approximation the actual spatially resolved concentration c _j (r) of the substances sought again: c * j ( r ) ≈ c j ( r )

Claims (13)

Computed tomographic image recording method in which, under irradiation of an examination subject ( 10 ) with polychromatic X-radiation (R) along several projection vectors ( d ) projection images (P) of the object to be examined ( 10 ) and in which by means of a reconstruction algorithm on the basis of the projection images (P) a location-dependent attenuation coefficient (κ ( r , E _i )) of the examination subject (FIG. 10 ), wherein the transmitted X-ray radiation (R) is detected in an energy-resolved manner, characterized in that a spatially dependent spectral attenuation coefficient (κ ( r , E _i )) is calculated by separate reconstruction of quasi-monoenergetic groups of projection images (P _i ). Image recording method according to claim 1, characterized in that on the basis of the spectral attenuation coefficient (κ ( r , E _i )) by weighted integration over the energy (E), a location-dependent energy-averaged attenuation coefficient (μ ( r )) is calculated. Image recording method according to claim 2, characterized in that the location-dependent spectral attenuation coefficient (κ ( r , E _i )) in the integration with the negative cube of the energy (E) is weighted. Method for determining the location-dependent concentration (c _j ( r )) of a number of predefined substances in an examination subject ( 10 ), characterized in that by means of a regression algorithm a model function (F ( r , E _i )) to a location-dependent spectral attenuation coefficient (κ ( r , E _i )) of the examination subject ( 10 ), wherein the model function (F ( r , E _i )) contains as fixed parameters the attenuation spectra (λ (E _i )) of the substances sought and as free parameters the location-dependent concentration (c _j ( r )) of these substances. A method according to claim 4, characterized in that a linear combination of the attenuation spectra (λ (E _i )) is used as a model function (F ( r , E _i )). Method according to claim 4 or 5, characterized in that the model function (F ( r , E _i )) contains the attenuation spectrum (λ (E _i )) of at least one element, in particular hydrogen, oxygen and / or calcium, as a fixed parameter. Method according to one of claims 4 to 6, characterized in that the model function (F ( r , E _i )) the attenuation spectrum (λ (E _i )) mindes least one homogeneous body substance, in particular water, fat, soft tissue or bone contains. Method according to one of claims 4 to 7, characterized in that the location-dependent spectral attenuation coefficient (κ ( r , E _i )) with the image recording method according to one of claims 1 to 3 is determined. Computer tomograph ( 1 ) for carrying out the image recording method according to one of claims 1 to 3, with a polychromatic X-ray source ( 3 ) and arranged in juxtaposition to this X-ray detector ( 4 ), wherein the X-ray source ( 3 ) and the X-ray detector ( 4 ) around a common isocentric axis ( 7 ) are rotatably arranged to be irradiated by an X-ray source ( 3 ) and X-ray detector ( 4 ) ( 10 ) along different projection vectors ( d ) to produce projection images (P, P _i ) of the transmitted X-ray radiation (R) and to a reconstruction unit ( 8th ) for calculating a location-dependent attenuation coefficient (κ ( r , E _i )) of the examination object ( 10 ) based on the projection images (P), wherein the X-ray detector ( 4 ) is designed for the energy-resolved detection of the transmitted X-radiation (R), characterized in that the reconstruction unit ( 8th ) is designed to calculate a spatially dependent spectral attenuation coefficients (κ ( r , E _i )) by respectively processing quasi-monoenergetic groups of projection images (P _i ) separately. Computer tomograph ( 1 ) according to claim 9, characterized in that the detector ( 4 ) is designed as a direct converter with an energy-resolved quantum counting electronics. Computer tomograph ( 1 ) according to claim 10, characterized in that the detector ( 4 ) comprises at least one scintillator layer with a downstream photosensor and a quantum counting electronics. Computer tomograph ( 1 ) according to claim 11, characterized in that the detector ( 4 ) has a sandwich arrangement of scintillator layers for energy-resolved detection of the transmitted X-radiation (R). Computer tomograph ( 1 ) according to one of claims 9 to 12, characterized by a concentration determination module ( 9 ), which is designed for carrying out the method according to one of claims 4 to 7.